Article Guide
How the Article Guide Works
Every article we publish opens with a short, plain-language companion: what the study asked, how it was done, what it found, and what it cannot tell you. This page explains how that companion is prepared and who is responsible for it.
What an Article Guide is
A research paper is written for specialists. The Article Guide restates the same study in clear English for anyone who needs to understand it quickly: a clinician between patients, a researcher outside the field, a journalist, a patient, a student.
It is a companion to the article, not a replacement for it. Nothing in a guide is new: every sentence describes something the published paper already says, and the paper itself remains the version of record.
How a guide is made
Three steps, in this order. A guide that has not completed all three never reaches the site.
-
A draft is written offline
A language model running on the editorial office's own computer reads the published article (title, abstract and full text) and drafts the summary. The model runs locally: the article is never sent to an external AI service, and the servers that host this website do not call one.
-
A journal editor checks every line
The editor handling the article compares each statement, each number, each stated limitation and each cited section against the paper, and rewrites anything that is imprecise, overstated or missing. The draft is a starting point for that editor, not an output to be waved through.
-
Nothing appears without approval
The guide becomes visible only when that editor approves it. Unapproved drafts are held inside the editorial system and are never published, and the approval date is shown on the guide itself.
What is inside every guide
Research question
What the study set out to answer.
Methods at a glance
The design, the population and how the outcome was measured.
Main findings
The principal results, with the numbers that carry them.
Limitations
What weakens the findings, stated plainly.
What the study does not establish
The claims the paper is often assumed to support but does not: usually causation, clinical recommendations, or generalisation beyond the population studied. This section exists because it is the one readers most often get wrong.
Where each answer comes from
Each question in a guide names the part of the article its answer was drawn from: the abstract, a named section, or a specific figure or table. Those labels are not free text: only headings and captions that actually exist in that article can be used, and an editor cannot approve a guide that cites anything else.
So a claim in a guide can always be traced back to a place in the paper where you can check it.
“Ask this article”
Each guide includes a question box. It is deliberately narrow in what it will do: it returns an editor-approved answer when your question matches one, and otherwise shows passages copied verbatim from the article itself.
It does not write new sentences, does not consult any external service, and does not keep a record of what you asked. A question the article does not address returns nothing rather than the nearest loosely-matching paragraph. An empty answer is more honest than a confident irrelevant one.
Kept in step with the article
If the title, abstract or full text of an article is corrected after its guide was approved, the guide is withdrawn from the article page automatically. It stays hidden until an editor has re-read it against the corrected text and approved it again.
A guide on the site therefore always describes the version of the article you are currently reading.
What an Article Guide is not
- Not medical advice. Nothing in a guide is guidance for treating a patient or for your own care. Speak to a qualified health professional.
- Not a substitute for the article. A summary drops detail by design. Anything you intend to cite, teach or act on should be read in the paper.
- Not part of peer review. A guide is written after acceptance and adds no assessment of the work. Peer review happens before publication and is unaffected by it.
Found something wrong?
Editors are responsible for what a guide says, so errors in one are ours to fix. If a guide misstates the study it accompanies, tell us which article it belongs to and what is wrong, and we will correct or withdraw it.